Statistical Analysis of Crop Yield Under Controlled Line‐source Irrigation

The line‐source sprinkler irrigation system produces a water application pattern which is uniform along the length of the plot and variable across the plot. This ensures a high degree of uniformity of deterministic water application along strips parallel to the sprinkler line and minimizes effect of irrigation variability upon yield in such strips. Statistical techniques are applied to analyze yield results obtained from controlled line‐source field experiments. The spatial statistical structure of the crop yield is determined. The completely random component of yield is separated from the one displaying a spatial structure attributed to soil properties. Two of the three experimental fields considered display the complete random component solely, so that the usual statistical inference techniques are applicable. In one field the crop yield has a spatial statistical structure and the experimental plot seems to be smaller than the correlation scale, indicating that yield average and variance estimates depend on the size of the field. The main conclusions are: (i) statistical stationarity is achieved when yield is scaled with respect to its average for a constant irrigation level; (ii) soil variability has a large effect on yield; and (iii) the latter has a spatial structure which has to be inferred in order to carry out statistical analysis independent of field size.